Considerations for Statistics Gathering

CData Virtuality’s optimization subsystem can work with cost-based optimization. As a best practice, we recommend gathering statistics when tables with more than 100 million rows are involved, both for the 100 million rows tables and the small tables they are joined with. This way, dependent joins will be created when joining very large tables with very small tables. For all other cases, it is sufficient to use Merge Joins so that the additional computational effort of gathering statistics can be saved.